Re: [Tracker] Internationalization in SPARQL queries

[Martyn Russell <martyn lanedo com>]

On 22/02/10 11:03, Florent ThÃvenet wrote:
> Le lundi 22 fÃvrier 2010 Ã 08:21 +0000, Martyn Russell a Ãcrit :
> > Hi,
> Hi Martyn

Hej ;)

> > This is quite an interesting idea. Using a function like
> > tracker:LocalizedName(?city) wouldn't work unless we stored the data in
> > multiple translations. This would increase the data quite significantly
> > but I suppose we could use something like tracker:should_translate (or
> > some property to know which properties to handle).
> >
> > The other problem I see with this is, we can't simply call the gettext
> > functions to do the translations at run time. Translations are (as you
> > likely know) supplied by the translation teams and would have to be
> > there before we release Tracker. So I am not sure how run-time words
> > could be translated at this point?
> I think Michele's solution is very good, what's more it's "sparql
> official".
> (http://mail.gnome.org/archives/tracker-list/2010-February/msg00141.htm )
> About the increase of the data: every property doesn't need to be
> translated (I don't think we will store the plainTextContent of a 700
> pages pdf file :-) ) and with nowadays hard drive, storing two strings
> instead of one is not terrible, even on embedded devices.

Of course, hence why I said I agreed with the idea that we have some 
ontology to know which properties are translatable ;)

--
Regards,
Martyn